Integrative approaches that simultaneously model multi-omics data have gained increasing popularity because they provide holistic system biology views of multiple or all components in a biological system of interest. Canonical correlation analysis (CCA) is a correlation-based integrative method designed to extract latent features shared between multiple assays by finding the linear combinations of features-referred to as canonical variables (CVs)-within each assay that achieve maximal across-assay correlation. Although widely acknowledged as a powerful approach for multi-omics data, CCA has not been systematically applied to multi-omics data in large cohort studies, which has only recently become available. Here, we adapted sparse multiple CCA (SMCCA), a widely-used derivative of CCA, to proteomics and methylomics data from the Multi-Ethnic Study of Atherosclerosis (MESA) and Jackson Heart Study (JHS). To tackle challenges encountered when applying SMCCA to MESA and JHS, our adaptations include the incorporation of the Gram-Schmidt (GS) algorithm with SMCCA to improve orthogonality among CVs, and the development of Sparse Supervised Multiple CCA (SSMCCA) to allow supervised integration analysis for more than two assays. Effective application of SMCCA to the two real datasets reveals important findings. Applying our SMCCA-GS to MESA and JHS, we identified strong associations between blood cell counts and protein abundance, suggesting that adjustment of blood cell composition should be considered in protein-based association studies. Importantly, CVs obtained from two independent cohorts also demonstrate transferability across the cohorts. For example, proteomic CVs learned from JHS, when transferred to MESA, explain similar amounts of blood cell count phenotypic variance in MESA, explaining 39.0% ~ 50.0% variation in JHS and 38.9% ~ 49.1% in MESA. Similar transferability was observed for other omics-CV-trait pairs. This suggests that biologically meaningful and cohort-agnostic variation is captured by CVs. We anticipate that applying our SMCCA-GS and SSMCCA on various cohorts would help identify cohort-agnostic biologically meaningful relationships between multi-omics data and phenotypic traits.

Genome-wide association studies have identified numerous loci with allelic associations to Type 1 Diabetes (T1D) risk. Most disease-associated variants are enriched in regulatory sequences active in lymphoid cell types, suggesting that lymphocyte gene expression is altered in T1D. Here we assay gene expression between T1D cases and healthy controls in two autoimmunity-relevant lymphocyte cell types, memory CD4+/CD25+ regulatory T cells (Treg) and memory CD4+/CD25- T cells, using a splicing event-based approach to characterize tissue-specific transcriptomes. Limited differences in isoform usage between T1D cases and controls are observed in memory CD4+/CD25- T-cells. In Tregs, 402 genes demonstrate differences in isoform usage between cases and controls, particularly RNA recognition and splicing factor genes. Many of these genes are regulated by the variable inclusion of exons that can trigger nonsense mediated decay. Our results suggest that dysregulation of gene expression, through shifts in alternative splicing in Tregs, contributes to T1D pathophysiology.

Mitochondria carry their own circular genome and disruption of the mitochondrial genome is associated with various aging-related diseases. Unlike the nuclear genome, mitochondrial DNA (mtDNA) can be present at 1000 s to 10,000 s copies in somatic cells and variants may exist in a state of heteroplasmy, where only a fraction of the DNA molecules harbors a particular variant. We quantify mtDNA heteroplasmy in 194,871 participants in the UK Biobank and find that heteroplasmy is associated with a 1.5-fold increased risk of all-cause mortality. Additionally, we functionally characterize mtDNA single nucleotide variants (SNVs) using a constraint-based score, mitochondrial local constraint score sum (MSS) and find it associated with all-cause mortality, and with the prevalence and incidence of cancer and cancer-related mortality, particularly leukemia. These results indicate that mitochondria may have a functional role in certain cancers, and mitochondrial heteroplasmic SNVs may serve as a prognostic marker for cancer, especially for leukemia.

Although many novel gene-metabolite and gene-protein associations have been identified using high-throughput biochemical profiling, systematic studies that leverage human genetics to illuminate causal relationships between circulating proteins and metabolites are lacking. Here, we performed protein-metabolite association studies in 3,626 plasma samples from three human cohorts. We detected 171,800 significant protein-metabolite pairwise correlations between 1,265 proteins and 365 metabolites, including established relationships in metabolic and signaling pathways such as the protein thyroxine-binding globulin and the metabolite thyroxine, as well as thousands of new findings. In Mendelian randomization (MR) analyses, we identified putative causal protein-to-metabolite associations. We experimentally validated top MR associations in proof-of-concept plasma metabolomics studies in three murine knockout strains of key protein regulators. These analyses identified previously unrecognized associations between bioactive proteins and metabolites in human plasma. We provide publicly available data to be leveraged for studies in human metabolism and disease.

We assess performance and limitations of polygenic risk scores (PRSs) for multiple blood pressure (BP) phenotypes in diverse population groups. We compare "clumping-and-thresholding" (PRSice2) and LD-based (LDPred2) methods to construct PRSs from each of multiple GWAS, as well as multi-PRS approaches that sum PRSs with and without weights, including PRS-CSx. We use datasets from the MGB Biobank, TOPMed study, UK biobank, and from All of Us to train, assess, and validate PRSs in groups defined by self-reported race/ethnic background (Asian, Black, Hispanic/Latino, and White). For both SBP and DBP, the PRS-CSx based PRS, constructed as a weighted sum of PRSs developed from multiple independent GWAS, perform best across all race/ethnic backgrounds. Stratified analysis in All of Us shows that PRSs are better predictive of BP in females compared to males, individuals without obesity, and middle-aged (40-60 years) compared to older and younger individuals.

Type 1 diabetes (T1D) is a chronic condition caused by autoimmune destruction of the insulin-producing pancreatic β cells. While it is known that gene-environment interactions play a key role in triggering the autoimmune process leading to T1D, the pathogenic mechanism leading to the appearance of islet autoantibodies-biomarkers of autoimmunity-is poorly understood. Here we show that disruption of the complement system precedes the detection of islet autoantibodies and persists through disease onset. Our results suggest that children who exhibit islet autoimmunity and progress to clinical T1D have lower complement protein levels relative to those who do not progress within a similar time frame. Thus, the complement pathway, an understudied mechanistic and therapeutic target in T1D, merits increased attention for use as protein biomarkers of prediction and potentially prevention of T1D.

We rigorously assessed a comprehensive association testing framework for heteroplasmy, employing both simulated and real-world data. This framework employed a variant allele fraction (VAF) threshold and harnessed multiple gene-based tests for robust identification and association testing of heteroplasmy. Our simulation studies demonstrated that gene-based tests maintained an appropriate type I error rate at α=0.001. Notably, when 5% or more heteroplasmic variants within a target region were linked to an outcome, burden-extension tests (including the adaptive burden test, variable threshold burden test, and z-score weighting burden test) outperformed the sequence kernel association test (SKAT) and the original burden test. Applying this framework, we conducted association analyses on whole-blood derived heteroplasmy in 17,507 individuals of African and European ancestries (31% of African Ancestry, mean age of 62, with 58% women) with whole genome sequencing data. We performed both cohort- and ancestry-specific association analyses, followed by meta-analysis on both pooled samples and within each ancestry group. Our results suggest that mtDNA-encoded genes/regions are likely to exhibit varying rates in somatic aging, with the notably strong associations observed between heteroplasmy in the RNR1 and RNR2 genes (p<0.001) and advance aging by the Original Burden test. In contrast, SKAT identified significant associations (p<0.001) between diabetes and the aggregated effects of heteroplasmy in several protein-coding genes. Further research is warranted to validate these findings. In summary, our proposed statistical framework represents a valuable tool for facilitating association testing of heteroplasmy with disease traits in large human populations.

Genome-wide association studies (GWAS) are widely applied to identify susceptibility loci for a variety of diseases using genotyping arrays that interrogate known polymorphisms throughout the genome. A particular strength of GWAS is that it is unbiased with respect to specific genomic elements (e.g., coding or regulatory regions of genes), and it has revealed important associations that would have never been suspected based on prior knowledge or assumptions. To date, the discovered SNPs associated with complex human traits tend to have small effect sizes, requiring very large sample sizes to achieve robust statistical power. To address these issues, a number of efficient strategies have emerged for conducting GWAS, including combining study results across multiple studies using meta-analysis, collecting cases through electronic health records, and using samples collected from other studies as controls that have already been genotyped and made publicly available (e.g., through deposition of de-identified data into dbGaP or EGA). In certain scenarios, it may be attractive to use already genotyped controls and divert resources to standardized collection, phenotyping, and genotyping of cases only. This strategy, however, requires that careful attention be paid to the choice of "public controls" and to the comparability of genetic data between cases and the public controls to ensure that any allele frequency differences observed between groups is attributable to locus-specific effects rather than to a systematic bias due to poor matching (population stratification) or differential genotype calling (batch effects). The goal of this paper is to describe some of the potential pitfalls in using previously genotyped control data. We focus on considerations related to the choice of control groups, the use of different genotyping platforms, and approaches to deal with population stratification when cases and controls are genotyped across different platforms.

Islet autoantibodies detected at disease onset in patients with type 1 diabetes are signs of an autoimmune destruction of the insulin-producing β-cells. To further investigate the genetic determinants of autoantibody positivity, we performed dense immune-focused genotyping on the Immunochip array and tested for association with seven disease-specific autoantibodies in a large cross-sectional cohort of 6,160 type 1 diabetes-affected siblings. The genetic association with positivity for GAD autoantibodies (GADAs), IA2 antigen (IA-2A), zinc transporter 8, thyroid peroxidase, gastric parietal cells (PCAs), tissue transglutaminase, and 21-hydroxylase was tested using a linear mixed-model regression approach to simultaneously control for population structure and family relatedness. Four loci were associated with autoantibody positivity at genome-wide significance. Positivity for GADA was associated with 3q28/LPP, for IA-2A with 1q23/FCRL3 and 11q13/RELA, and for PCAs with 2q24/IFIH1. The 3q28 locus showed association after only 3 years duration and might therefore be a marker of persistent GADA positivity. The 1q23, 11q13, and 2q24 loci were associated with autoantibodies close to diabetes onset and constitute candidates for early screening. Major susceptibility loci for islet autoantibodies are separate from type 1 diabetes risk, which may have consequences for intervention strategies to reduce autoimmunity.

The Environmental Determinants of Diabetes in the Young (TEDDY) study prospectively follows 8,677 children enrolled from birth who carry HLA-susceptibility genotypes for development of islet autoantibodies (IA) and type 1 diabetes (T1D). During the median follow-up time of 57 months, 350 children developed at least one persistent IA (GAD antibody, IA-2A, or micro insulin autoantibodies) and 84 of them progressed to T1D. We genotyped 5,164 Caucasian children for 41 non-HLA single nucleotide polymorphisms (SNPs) that achieved genome-wide significance for association with T1D in the genome-wide association scan meta-analysis conducted by the Type 1 Diabetes Genetics Consortium. In TEDDY participants carrying high-risk HLA genotypes, eight SNPs achieved significant association to development of IA using time-to-event analysis (P < 0.05), whereof four were significant after adjustment for multiple testing (P < 0.0012): rs2476601 in PTPN22 (hazard ratio [HR] 1.54 [95% CI 1.27-1.88]), rs2292239 in ERBB3 (HR 1.33 [95% CI 1.14-1.55]), rs3184504 in SH2B3 (HR 1.38 [95% CI 1.19-1.61]), and rs1004446 in INS (HR 0.77 [0.66-0.90]). These SNPs were also significantly associated with T1D in particular: rs2476601 (HR 2.42 [95% CI 1.70-3.44]). Although genes in the HLA region remain the most important genetic risk factors for T1D, other non-HLA genetic factors contribute to IA, a first step in the pathogenesis of T1D, and the progression of the disease.

Fasting glucose and insulin are intermediate traits for type 2 diabetes. Here we explore the role of coding variation on these traits by analysis of variants on the HumanExome BeadChip in 60,564 non-diabetic individuals and in 16,491 T2D cases and 81,877 controls. We identify a novel association of a low-frequency nonsynonymous SNV in GLP1R (A316T; rs10305492; MAF=1.4%) with lower FG (β=-0.09±0.01 mmol l(-1), P=3.4 × 10(-12)), T2D risk (OR[95%CI]=0.86[0.76-0.96], P=0.010), early insulin secretion (β=-0.07±0.035 pmolinsulin mmolglucose(-1), P=0.048), but higher 2-h glucose (β=0.16±0.05 mmol l(-1), P=4.3 × 10(-4)). We identify a gene-based association with FG at G6PC2 (pSKAT=6.8 × 10(-6)) driven by four rare protein-coding SNVs (H177Y, Y207S, R283X and S324P). We identify rs651007 (MAF=20%) in the first intron of ABO at the putative promoter of an antisense lncRNA, associating with higher FG (β=0.02±0.004 mmol l(-1), P=1.3 × 10(-8)). Our approach identifies novel coding variant associations and extends the allelic spectrum of variation underlying diabetes-related quantitative traits and T2D susceptibility.

So far, more than 170 loci have been associated with circulating lipid levels through genome-wide association studies (GWAS). These associations are largely driven by common variants, their function is often not known, and many are likely to be markers for the causal variants. In this study we aimed to identify more new rare and low-frequency functional variants associated with circulating lipid levels.

Despite finding more than 40 risk loci for type 1 diabetes (T1D), the causative variants and genes remain largely unknown. Here, we sought to identify rare deleterious variants of moderate-to-large effects contributing to T1D. We deeply sequenced 301 protein-coding genes located in 49 previously reported T1D risk loci in 70 T1D cases of European ancestry. These cases were selected from putatively high-risk families that had three or more siblings diagnosed with T1D at early ages. A cluster of rare deleterious variants in PTPN22 was identified, including two novel frameshift mutations (ss538819444 and rs371865329) and two missense variants (rs74163663 and rs56048322). Genotyping in 3,609 T1D families showed that rs56048322 was significantly associated with T1D and that this association was independent of the T1D-associated common variant rs2476601. The risk allele at rs56048322 affects splicing of PTPN22, resulting in the production of two alternative PTPN22 transcripts and a novel isoform of LYP (the protein encoded by PTPN22). This isoform competes with the wild-type LYP for binding to CSK and results in hyporesponsiveness of CD4(+) T cells to antigen stimulation in T1D subjects. These findings demonstrate that in addition to common variants, rare deleterious variants in PTPN22 exist and can affect T1D risk.

Homozygosity has long been associated with rare, often devastating, Mendelian disorders, and Darwin was one of the first to recognize that inbreeding reduces evolutionary fitness. However, the effect of the more distant parental relatedness that is common in modern human populations is less well understood. Genomic data now allow us to investigate the effects of homozygosity on traits of public health importance by observing contiguous homozygous segments (runs of homozygosity), which are inferred to be homozygous along their complete length. Given the low levels of genome-wide homozygosity prevalent in most human populations, information is required on very large numbers of people to provide sufficient power. Here we use runs of homozygosity to study 16 health-related quantitative traits in 354,224 individuals from 102 cohorts, and find statistically significant associations between summed runs of homozygosity and four complex traits: height, forced expiratory lung volume in one second, general cognitive ability and educational attainment (P < 1 × 10(-300), 2.1 × 10(-6), 2.5 × 10(-10) and 1.8 × 10(-10), respectively). In each case, increased homozygosity was associated with decreased trait value, equivalent to the offspring of first cousins being 1.2 cm shorter and having 10 months' less education. Similar effect sizes were found across four continental groups and populations with different degrees of genome-wide homozygosity, providing evidence that homozygosity, rather than confounding, directly contributes to phenotypic variance. Contrary to earlier reports in substantially smaller samples, no evidence was seen of an influence of genome-wide homozygosity on blood pressure and low density lipoprotein cholesterol, or ten other cardio-metabolic traits. Since directional dominance is predicted for traits under directional evolutionary selection, this study provides evidence that increased stature and cognitive function have been positively selected in human evolution, whereas many important risk factors for late-onset complex diseases may not have been.

There is growing evidence of shared risk alleles for complex traits (pleiotropy), including autoimmune and neuropsychiatric diseases. This might be due to sharing among all individuals (whole-group pleiotropy) or a subset of individuals in a genetically heterogeneous cohort (subgroup heterogeneity). Here we describe the use of a well-powered statistic, BUHMBOX, to distinguish between those two situations using genotype data. We observed a shared genetic basis for 11 autoimmune diseases and type 1 diabetes (T1D; P < 1 × 10(-4)) and for 11 autoimmune diseases and rheumatoid arthritis (RA; P < 1 × 10(-3)). This sharing was not explained by subgroup heterogeneity (corrected PBUHMBOX > 0.2; 6,670 T1D cases and 7,279 RA cases). Genetic sharing between seronegative and seropostive RA (P < 1 × 10(-9)) had significant evidence of subgroup heterogeneity, suggesting a subgroup of seropositive-like cases within seronegative cases (PBUHMBOX = 0.008; 2,406 seronegative RA cases). We also observed a shared genetic basis for major depressive disorder (MDD) and schizophrenia (P < 1 × 10(-4)) that was not explained by subgroup heterogeneity (PBUHMBOX = 0.28; 9,238 MDD cases).

Previously we performed a linkage scan of 638 African American affected sibling pairs (ASP) with type 2 diabetes (T2D) enriched for end-stage renal disease (ESRD). Ordered subset linkage analysis (OSA) revealed a linkage peak on chromosome 7p in the subset of families with earlier age of T2D diagnosis.

A substantial genetic contribution to systemic lupus erythematosus (SLE) risk is conferred by major histocompatibility complex (MHC) gene(s) on chromosome 6p21. Previous studies in SLE have lacked statistical power and genetic resolution to fully define MHC influences. We characterized 1,610 Caucasian SLE cases and 1,470 parents for 1,974 MHC SNPs, the highly polymorphic HLA-DRB1 locus, and a panel of ancestry informative markers. Single-marker analyses revealed strong signals for SNPs within several MHC regions, as well as with HLA-DRB1 (global p = 9.99 x 10(-16)). The most strongly associated DRB1 alleles were: *0301 (odds ratio, OR = 2.21, p = 2.53 x 10(-12)), *1401 (OR = 0.50, p = 0.0002), and *1501 (OR = 1.39, p = 0.0032). The MHC region SNP demonstrating the strongest evidence of association with SLE was rs3117103, with OR = 2.44 and p = 2.80 x 10(-13). Conditional haplotype and stepwise logistic regression analyses identified strong evidence for association between SLE and the extended class I, class I, class III, class II, and the extended class II MHC regions. Sequential removal of SLE-associated DRB1 haplotypes revealed independent effects due to variation within OR2H2 (extended class I, rs362521, p = 0.006), CREBL1 (class III, rs8283, p = 0.01), and DQB2 (class II, rs7769979, p = 0.003, and rs10947345, p = 0.0004). Further, conditional haplotype analyses demonstrated that variation within MICB (class I, rs3828903, p = 0.006) also contributes to SLE risk independent of HLA-DRB1*0301. Our results for the first time delineate with high resolution several MHC regions with independent contributions to SLE risk. We provide a list of candidate variants based on biologic and functional considerations that may be causally related to SLE risk and warrant further investigation.

We conducted a genome-wide association study of rheumatoid arthritis in 2,418 cases and 4,504 controls from North America and identified an association at the REL locus, encoding c-Rel, on chromosome 2p13 (rs13031237, P = 6.01 x 10(-10)). Replication in independent case-control datasets comprising 2,604 cases and 2,882 controls confirmed this association, yielding an allelic OR = 1.25 (P = 3.08 x 10(-14)) for marker rs13031237 and an allelic OR = 1.21 (P = 2.60 x 10(-11)) for marker rs13017599 in the combined dataset. The combined dataset also provides definitive support for associations at both CTLA4 (rs231735; OR = 0.85; P = 6.25 x 10(-9)) and BLK (rs2736340; OR = 1.19; P = 5.69 x 10(-9)). c-Rel is an NF-kappaB family member with distinct functional properties in hematopoietic cells, and its association with rheumatoid arthritis suggests disease pathways that involve other recently identified rheumatoid arthritis susceptibility genes including CD40, TRAF1, TNFAIP3 and PRKCQ.

Mutations within the NOTCH3 gene cause cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). CADASIL mutations appear to be restricted to the first twenty-four exons, resulting in the gain or loss of a cysteine amino acid. The role of other exonic NOTCH3 variation not involving cysteine residues and mutations in exons 25-33 in ischemic stroke remains unresolved.

Most common human traits and diseases have a polygenic pattern of inheritance: DNA sequence variants at many genetic loci influence the phenotype. Genome-wide association (GWA) studies have identified more than 600 variants associated with human traits, but these typically explain small fractions of phenotypic variation, raising questions about the use of further studies. Here, using 183,727 individuals, we show that hundreds of genetic variants, in at least 180 loci, influence adult height, a highly heritable and classic polygenic trait. The large number of loci reveals patterns with important implications for genetic studies of common human diseases and traits. First, the 180 loci are not random, but instead are enriched for genes that are connected in biological pathways (P = 0.016) and that underlie skeletal growth defects (P < 0.001). Second, the likely causal gene is often located near the most strongly associated variant: in 13 of 21 loci containing a known skeletal growth gene, that gene was closest to the associated variant. Third, at least 19 loci have multiple independently associated variants, suggesting that allelic heterogeneity is a frequent feature of polygenic traits, that comprehensive explorations of already-discovered loci should discover additional variants and that an appreciable fraction of associated loci may have been identified. Fourth, associated variants are enriched for likely functional effects on genes, being over-represented among variants that alter amino-acid structure of proteins and expression levels of nearby genes. Our data explain approximately 10% of the phenotypic variation in height, and we estimate that unidentified common variants of similar effect sizes would increase this figure to approximately 16% of phenotypic variation (approximately 20% of heritable variation). Although additional approaches are needed to dissect the genetic architecture of polygenic human traits fully, our findings indicate that GWA studies can identify large numbers of loci that implicate biologically relevant genes and pathways.